lp:~lttng/lttng-modules/trunk

Branch merges

Related source package recipes

1 recipe using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Related blueprints

1233. By Jonathan Rajotte on 2019-08-13

Introduce LTTNG_KERNEL_SESSION_SET_CREATION_TIME

Add trace_creation_datetime to the metadata env field

This allows a viewer to get more information regarding the creation time
of a trace. This information is normally inferred from the trace
hierarchy.

The ABI expects an ISO8601 compliant string value. We leave the
formatting to the trace controller.

Signed-off-by: Jonathan Rajotte <email address hidden>
Signed-off-by: Mathieu Desnoyers <email address hidden>

1232. By Jonathan Rajotte on 2019-08-13

Add metadata env fields

Add the following fields:

  - tracer_buffering_scheme
      The buffering scheme used by the tracer. lttng-modules sole
      buffering scheme is "global".
  - trace_name
      The name of the trace. Use the session name.

Reviewed-by: Mathieu Desnoyers <email address hidden>
Signed-off-by: Jonathan Rajotte <email address hidden>
Signed-off-by: Mathieu Desnoyers <email address hidden>

1231. By Jonathan Rajotte on 2019-08-13

Introduce LTTNG_KERNEL_SESSION_SET_NAME

The tracer controller (lttng-sessiond) can now inform the kernel tracer
of the name of the created session. This will allow the tracer to
propagate the information inside the trace metadata under a "env" field.

This information is normally inferred from the generated folder
structure where a trace rests.

Reviewed-by: Mathieu Desnoyers <email address hidden>
Signed-off-by: Jonathan Rajotte <email address hidden>
Signed-off-by: Mathieu Desnoyers <email address hidden>

1230. By Jonathan Rajotte on 2019-07-04

Fix: do not use diagnostic pragma when GCC version is lower than 4.6.0

Officially the diagnostic pragma are supported starting in 4.6. [1]
But they were present before 4.6 with limitation which we cannot
honour easily.

[1] https://gcc.gnu.org/gcc-4.6/changes.html

Signed-off-by: Jonathan Rajotte <email address hidden>
Signed-off-by: Mathieu Desnoyers <email address hidden>

1229. By Jonathan Rajotte on 2019-07-04

Fix: missing define when not building with gcc

Signed-off-by: Jonathan Rajotte <email address hidden>
Signed-off-by: Mathieu Desnoyers <email address hidden>

1228. By Mathieu Desnoyers on 2019-06-24

Fix: lttng-tracepoint module notifier should return NOTIFY_OK

Module notifiers should return NOTIFY_OK on success rather than the
value 0. The return value 0 does not seem to have any ill side-effects
in the notifier chain caller, but it is preferable to respect the API
requirements in case this changes in the future.

Notifiers can encapsulate a negative errno value with
notifier_from_errno(), but this is not needed by the LTTng tracepoint
notifier.

The approach taken in this notifier is to just print a console warning
on error, because tracing failure should not prevent loading a module.
So we definitely do not want to stop notifier iteration. Returning
an error without stopping iteration is not really that useful, because
only the return value of the last callback is returned to notifier chain
caller.

Signed-off-by: Mathieu Desnoyers <email address hidden>

1227. By Jérémie Galarneau <email address hidden> on 2019-06-11

Fix: Don't print ring-buffer's records count when it is not used

The teardown of a ring buffer causes a number of diagnostic messages
to be printed using printk. One of those contains the "records
count", which is only updated when lttng-modules is built with
LTTNG_RING_BUFFER_COUNT_EVENTS defined.

Move the "records count" printing to a different function and stub it
out when LTTNG_RING_BUFFER_COUNT_EVENTS is not defined
(default configuration).

This eliminates messages of the following form from the dmesg output
when an LTTng session is torn down.

[...] ring buffer relay-discard, cpu 0: 0 records written, 0 records overrun

Signed-off-by: Jérémie Galarneau <email address hidden>
Signed-off-by: Mathieu Desnoyers <email address hidden>

1226. By Mathieu Desnoyers on 2019-06-11

Fix: do not set quiescent state on channel destroy

Setting the quiescent state to true for each stream at channel
destruction is not useful: there are no readers left anyway at
that stage.

The side-effect perceived of setting this quiescent state on
destroy is that the metadata stream ends up with an empty last
packet (due to flush_empty performed when setting the quiescent state)
which is never consumed. This shows up in the lttng-modules error
reporting.

Signed-off-by: Mathieu Desnoyers <email address hidden>

1225. By Mathieu Desnoyers on 2019-06-04

Fix: ring_buffer_frontend.c: init read timer with uninitialized flags

For the config->alloc RING_BUFFER_ALLOC_GLOBAL (metadata channel), the
read timer flags argument is uninitialized.

Found by Coverity:
CID 1401114 (#1 of 1): Uninitialized scalar variable (UNINIT)
6. uninit_use_in_call: Using uninitialized value flags when calling init_timer_key.

Signed-off-by: Mathieu Desnoyers <email address hidden>

1224. By Mathieu Desnoyers on 2019-05-22

Introduce callstack stackwalk implementation header

Introduce a new implementation header for the stackwalk-based API, added
in Linux 5.2 and gradually integrated within each architecture.

Signed-off-by: Mathieu Desnoyers <email address hidden>